THE IONIZED STELLAR WIND IN VELA X-1 DURING ECLIPSE Norbert S. Schulz, 1 Claude R. Canizares, 1 Julia C. Lee, 1 and Masao Sako 2

Transcription

1 The Astrophysical Journal, 64:L21 L2, 2002 January The American Astronomical Society. All rights reserved. Printed in U.S.A. THE IONIZED STELLAR WIND IN VELA X-1 DURING ECLIPSE Norbert S. Schulz, 1 Claude R. Canizares, 1 Julia C. Lee, 1 and Masao Sako 2 Received 2001 September 22; accepted 2001 November 28; published 2001 December 13 ABSTRACT We present a first analysis of a high-resolution X-ray spectrum of the ionized stellar wind of Vela X-1 during eclipse. The data were obtained with the High-Energy Transmission Grating Spectrometer on board the Chandra 4 X-Ray Observatory. The spectrum is resolved into emission lines with fluxes between 0.02 and 1.04 # photons cm s. We identify lines from a variety of charge states, including fluorescent lines from cold material and a warm photoionized wind. We can exclude signatures from collisionally ionized plasmas. For the first time, we identify fluorescent lines from L-shell ions from lower Z elements. We also detect radiative recombination continua from a kt p 10 ev ( 1.2 # 10 K) photoionized optically thin gas. The fluorescent line fluxes imly the existence of optically thick and clumped matter within or outside the warm photoionized plasma. Subject headings: binaries: eclipsing stars: early-type stars: neutron techniques: spectroscopic X-rays: stars 1. INTRODUCTION Massive early-type stars show substantial mass loss through a strong wind. Matter is radially accelerated via momentum transfer from the stellar UV radiation field through absorption and scattering in resonance lines of abundant ions (Castor, Abbott, & Klein 197). In general, the stellar wind itself is a source of X-rays, which form through reverse shocks generated by highdensity material in the wind (Lucy & White 1980; Owocki, Castor, & Rybicki 1988; Feldmeier et al. 1997). Typical observed X-ray luminosities range from a few times ergs s for earlytype B stars to a few times ergs s for massive O stars (Chlebowski, Harnden, & Sciortino 1989; Berghöfer, Schmitt, & Cassinelli 1996). Recent Chandra and XMM-Newton observations resolved the X-ray emission of several early-type stars, confirming the existence of collisionally ionized plasmas, but also suggested the possibility that alternative emission mechanisms for the observed X-ray emission need to be investigated (Schulz et al. 2000; Kahn et al. 2001; Waldron & Cassinelli 2001). The presence of a neutron star companion leads to a typical Bondi-Hoyle accretion flow, in which a fraction of the stellar wind gets swept up by the neutron star, which generates a strong X-ray continuum. The X-ray luminosity of such a source is many orders of magnitudes larger than the one expected from shocks in the stellar wind, and X-rays from that source are reprocessed in the wind, resulting in discrete emission lines (Stevens & Kallman 1990). A prototype of such a system is the high-mass X-ray binary Vela X-1 (4U ). Its early-type primary star HD 7781 is a massive B0. Ib type supergiant (Brucato & Kristian 1972) with a mass and radius of 34 R, and 23 M, (from Nagase 1989), respectively, and an inferred mass-loss rate of less than M, yr (Sato et al. 1986; Sako et al. 1999). The terminal 1 wind velocity has been determined to be 1700 km s (Dupree et al. 1980). The neutron star exhibits 282 s X-ray pulsations (Forman et al. 1973) and orbits the center of mass of the system at a distance of only about 0.6 stellar radii from the surface of 1 Center for Space Research, Massachusetts Institute of Technology, 70 Vassar Street, Building 37, Cambridge, MA 02139; nss@space.mit.edu, crc@space.mit.edu, jlee@space.mit.edu. 2 Columbia Astrophysics Laboratory and Department of Physics, Columbia University, 38 West 120th Street, New York, NY 10027; masao@astro.columbia.edu. L21 the supergiant ( a p 3.4 R, ) with a period of 8.96 days. This means that the neutron star is deeply embedded within the acceleration zone of the stellar wind. The X-ray continuum of Vela X-1 has been studied extensively in X-rays. The general shape can be explained by a power law with a high-energy cutoff typical for X-ray pulsars (White, Swank, & Holt 1983). Its average X-ray luminosity is ergs s, which is consistent with accretion from a stellar wind without Roche lobe overflow (Conti 1978). In this Letter, we are primarily concerned with X-ray line emission from Vela X-1 during the middle of its eclipse. A first analysis of a moderate-resolution X-ray line spectrum of Vela X-1 during eclipse was performed on a set observations with ASCA, which covered an entire eclipse transition from ingress to egress (Nagase et al. 1994). The observations revealed lines from He- and H-like ions species. It was suggested that the He-like transitions are formed by cascades following radiative recombination. Sako et al. (1999) identified recombination lines and radiative recombination continua (RRCs) produced by photoionization in an extended stellar wind. The deduced differential emission measure distribution was consistent with the wind model by Hatchett & McCray (1977). Sako et al. also indicated the existence of fluorescent lines from ion species of lower Z than Fe. 2. CHANDRA OBSERVATIONS AND DATA REDUCTION Vela X-1 was observed with the High-Energy Transmission Grating Spectrometer (HETGS) on 2000 April 13 (09:7:2 UT) continuously for 28 ks. The HETGS carries two different types of transmission gratings, the Medium-Energy Gratings (MEGs) and the High-Energy Gratings (HEGs). It allows for highresolution spectroscopy between about 1 and 3 Å with a peak spectral resolution at 12 Å of l/dl 1400 and at 1.8 Å of l/dl 180 in first-order HEG. The dispersed spectra were recorded with an array of six charge-coupled devices (CCDs). We refer to the available Chandra X-ray Center (CXC) documents for more detailed descriptions of the spectroscopic instruments. 3 We recorded a total of 3200 events in the co-added first-order MEG and 1700 events in the HEG after standard grade selection. The CXC provided aspect-corrected event lists via standard pipe- 3 See

2 L22 X-RAY EMISSION LINES FROM VELA X-1 Vol. 64 Fig. 1. Residual HETGS spectrum in 0.02 Å bin with identified lines. The fluorescent lines are plotted in green. The highly ionized lines (blue) and RRC features (red) indicate an optically thin photoionized plasma. line processing. These data were reprocessed using the latest available data processing products. The wavelength scale has a zero-point accuracy of Å in MEG and Å in firstorder HEG. The current status of the overall wavelength calibration is better than 0.0%. For the line analysis of the highresolution spectra, we divide first by the instrument effective area and exposure, which is legitimate, because the spectral bin size is larger than the actual response of the instrument. The image in the zeroth order provides a source position as well as a medium-resolution X-ray spectrum. The position of HD 7781 is well known to a precision of about 0.3 (Röser, Bastian, & Wiese 1991). After correcting for a 2 aspect drift, we find the position of the X-ray source to be within 0. of this position. We also find no other X-ray sources within a radius of 30 around Vela X-1 with a flux above 1.2 # ergs cm s. The CCD spectrum of that image also allows us to draw comparisons with previous ASCA observations, which are of similar spectral resolution but cover a larger fraction of the eclipse. This CCD spectrum is only moderately piled up (7%) and can be used to determine qualitative characteristics of the observed emission as well as the total observed luminosity. The continuum model is the same as used by Sako et al. (1999), which consists of two absorbed power laws from the scattered component and a direct component with indices fixed to 1.7. The choice of this model is only motivated by the fact that we can draw comparisons with the ASCA results specifically. Since the spectrum shows significant line emission, we added several Gaussian functions to fit the brighter lines observed in the ASCA spectrum (see Table 2 in Sako et al. 1999). In general, we find good agreement with the ASCA results. The most striking difference is the column density of the direct component, which is twice as high and also indicates a much fainter direct component. This is a consequence of the fact that the Chandra observations exposure covers a smaller range in orbital phase around phase zero. We verified this by reanalyzing the ASCA data using the Chandra phase interval. From an X ray flux of 0.88 # 10 ergs cm s, we then deduce # 10 ergs s for the luminosity (0. 10 kev), assuming a distance of 1.9 kpc (Sadakane et al. 198). 3. THE HETGS SPECTRA The spectral continuum is well modeled in the zero-order spectrum, and we use it to subtract the continuum from the summed and background-subtracted first-order grating data. Figure 1 shows the residual spectrum between 3.1 and 12. Å Fluorescent Emission The spectrum in Figure 1 shows a variety of fluorescent lines (green model) from Mg, Si, S, Ar, Ca, and Fe. The line properties, including possible identifications, are listed in Table 1. For identifications, we use the adjusted calculations from House (1969). A few are also tabulated in Mewe (1994). The measured lines do not show significant velocity shifts or broadening and can be generally identified with near neutral charge states. For the elements Fe, Ca, and Ar, we cannot distinguish between different charge states, and here Table 1 lists the most likely range based on the wavelengths given in House (1969). In the case of Mg, Si, and S, we can identify various states; however, for the lowest charge state, we can again only identify a range of most likely ionization states. For each detected line, we calculate an equivalent hydrogen column density required to produce the observed line fluxes

3 No. 1, 2002 SCHULZ ET AL. L23 TABLE 1 Fluorescent Line Energies and Fluxes Ion Type a (Å) a l exp l meas (Å) Flux b c Y eff Fe... Kb Ka ii xi Ca... ii vii Ar... vi ix S... ix iv viii Si... ix viii vii vi ii vi Mg... v ii iv a From House b In units of photons cm 2 s c Fraction of column density with respect to Fe K (uncertainties are 1%, in case of Mg 4%). using the method described in Sako et al. (1999). Fluorescent yields and photoionization cross sections are adopted from Kaastra & Mewe (1993) and Verner et al. (1996), respectively. Kaastra & Mewe (1993) predict an Fe Kb/Ka flux ratio of We fix the width of the Kb line to that of the Ka line and infer a ratio of Since this is consistent with the theoretical value, we assume that the iron line flux, and hence the estimated column density, is not affected by resonant Auger destruction (Ross, Fabian, & Brandt 1996). Resonant Auger destruction occurs in regions where line optical depths are significant. A fluorescent photon from the K shell can be resonantly absorbed by ions that have vacancies in the L shell. However, due to high-autoionization yields in atoms of much lower Z than Fe, the L-shell ion will most likely Auger-decay. In Table 1, the quantity Y eff denotes the fraction of the Fe Ka column density for each identified charge state of the other elements. For Fe Ka, we derive a column density of 3.3 # cm. If the Y eff s are added up for each element, we observe a correlation of the column density with Z, which could indicate the process of resonant Auger destruction at work. Such an interpretation was also given by Sako et al. (1999). However, for Si, such an interpretation is problematic since the Si vii and Si viii lines appear exceptionally strong braking that correlation. In this respect, the detection of fluorescent lines from L-shell ions suggests a somewhat modified explanation for the observed behavior Radiative Recombination Continua In photoionized plasmas, free electrons are captured by ions, which result in a continuous emission feature above the recombination edge (for a review, see Liedahl et al and references therein). Since the plasma is overionized relative to the local electron temperature, the widths of these continua are narrow and appear as linelike features in the spectrum. RRCs from Ne x and Ne ix at 9.10 and Å, respectively, are clearly observed. Although some of the features are rather weak, they are consistent with kt p 10 2 ev (for Ne x) or a temperature of T 1.2 # 10 K. The Mg xi RRC is blended with the Si fluorescent lines, and a possible RRC of O viii is observed at 14.2 Å. The detection of such narrow RRCs clearly indicates that there is a warm photoionized medium associated with the wind. For Ne x at and 9.10 Å, the measured Lya/RRC flux ratio is after correcting for interstellar column density and may be higher if there is additional absorption above the Galactic value. Liedahl & Paerels (1996) predict for the optically thin limit a ratio of 1.3 for a purely recombining plasma. Since this is inconsistent with the observed ratio, we assume that the assumption of a purely recombining plasma is not correct here. A more recent calculation by A. Kinkhabwala et al. (2001, in preparation) also includes photoexcitations in a more self-consistent manner and predicts a ratio of Lya/RRC p 2.7 at kt p 10 ev, which agrees well with our observations. The spectrum also exhibits very weak but detectable amounts of line flux from Fe L ions. This is consistent with the prediction by Kallman et al. (1996) that recombination emission from K- shell ions dominates the spectrum in an overionized plasma H- and He-like Lines The spectrum shown in Figure 2 also shows very strong emission lines (blue model) from H- and He-like S, Si, Mg, Ne, and possibly O. The measured line fluxes range between Fig. 2. Left: Fe K region, which includes Fe Ka and Kb line emission as well as a highly ionized Fe xxv line. Middle: Sixiii He-like triplet, which shows a dominating resonance and forbidden line. Right: Nex RRC, which appears at 9.10 Å below the Mg He-like triplet. Its fit indicates a temperature of 10 ev.

4 L24 X-RAY EMISSION LINES FROM VELA X-1 Vol and 3 # 10 photons cm 2 s 1 and appear unresolved with no significant systematic line shifts. In the case of Ne x, four lines from the Lyman series are detected, with Lya : b : g : d line ratios of 1.0 : 0.32 : 0.24 : 0.10 (with uncertainties on the order of 2%). The fluxes of higher lines in the Lyman series are larger than those expected from either a purely recombining plasma or a purely collisionally ionized plasma and implies the presence of photoexcitation. This is consistent with the high Lya/RRC flux ratio found in the previous section. For a photoionized plasma including the effects of photoexcitation selfconsistently, the theoretical ratios for Ne x at kt p 10 ev are 1.0 : 0.3 : 0.16 : 0.08 at a column density of NNe x p 3 # cm and a turbulent velocity of 300 km s 1 (A. Kinkhabwala et al. 2001, in preparation). The Lyman series for Mg xii is difficult to assess since the Lyb and Lyg appear blended. The He-like lines of Si xiii and Mg xi are shown in Figure 2 (middle and right). In both cases, the resonance and forbidden line fluxes have roughly equal strengths, which is not expected in a predominantly low density recombining plasma but is consistent with the fact that effects from photoexcitation have to be considered. It would therefore be incorrect to interpret these triplets using the calculations by Porquet & Dubau (2000), which in fact would predict a hot and dense collisional-dominated plasma. Such an interpretation would be in contrast with the results above, which indicate the presence of cool and warm photoionized media through the detection of a narrow RRC and a moderate line strength. Here Wojdowski et al. (2001) suggest that resonance scattering is the most probable process to enhance the resonance line in the He-like triplets. We have yet to include contributions from the companion wind itself. In this case, we have to consider collisionally ionized plasmas. Using the Astrophysical Plasma Emission Code (Smith et al. 2001), we can calculate line ratios between the Si xiii resonance line and Fe L lines at representative temperatures. Temperatures in O stars range between # 10 6 K(as measured in z Ori; Waldron & Cassinelli 2001) and, in an 7 extreme case, # 10 K (as measured in v Ori C; Schulz et al. 2000) from highly resolved X-ray lines. For now, we assume that this is also a representative range of possible temperatures in the winds of B0. Ib supergiants. In the high-temperature case, we can exclude a significant contribution because here Fe xxiv lines should dominate the spectrum, which we do not observe. Similarly, in the low-temperature case, Fe xvii lines should dominate the spectrum, which we do not observe either. For the intermediate range, we estimate, under the assumption that the sporadic and weak detections of Fe L lines indeed reflect the existence of collisionally ionized emission, a contribution of a few percent at most. 4. CONCLUSIONS The X-ray spectrum of Vela X-1 during eclipse shows a large variety of emission features, including lines from H-like and He-like ions, RRCs, a population of very weak Fe L lines, and a number of fluorescent emission lines from mid-z elements and iron indicating a complex structure of the X-ray emitting plasma. From the fit of the shape of the Ne x RRC at 9.10 Å, we deduce a temperature of 1.2 # 10 K for the plasma. Its flux ratio with the Ne x Lya line is consistent with an optically thin photoionized plasma. The spectrum shows clear evidence of photoexcitation processes, for which resonance scattering from the strong eclipsed X-ray continuum may be a likely source. More detailed calculations on this issue are required. Fluorescent lines from various charge states were detected and partially identified. They include strong emission from M-shell ions but also from various L-shell ions. Fluorescent yields are generally extremely low in mid-z elements, even at moderate optical depths. The correlation of the inferred column density with Z implies that autoionization dominates the fluorescent yields in these ions and that the fluorescent plasma is not optically thin. If the fluorescent plasma coexists with an optically thin photoionized plasma or in a region outside the warm photoionized region, its has to be in the form of dense cool clumps. Both scenarios may be the case, which is not inconsistent with current stellar wind models. For example, Feldmeier (199) pointed out that the wind can be divided into two distinct regions, an active inner one where shell-to-shell collisions occur and an outer, quiescent region with old material. The detection of fluorescence from L-shell ions is still mysterious. Given the relative low fluorescent yields for these ions, resonant Auger destruction would have almost completely destroyed these lines. Further investigations of the possible mechanisms and physical conditions that prevent the lines from being destroyed are required. The authors want to thank the MIT HETG team and the CXC for their support and Ali Kinkhabwala for providing results from calculations of photoionized plasmas in progress. This research is funded by contracts SV and NAS REFERENCES Berghöfer, T. W., Schmitt, J. H. M. M., & Cassinelli, J. P. 1996, A&AS, 118, 481 Brucato, R. J., & Kristian, J. 1972, ApJ, 173, L10 Castor, J. L., Abbott, D. C., & Klein, R. I. 197, ApJ, 19, 17 Chlebowski, T., Harnden, F. R., Jr., & Sciortino, S. 1989, ApJ, 341, 427 Conti, P. 1978, A&A, 63, 22 Dupree, A. K., et al. 1980, ApJ, 238, 969 Feldmeier, A. 199, A&A, 299, 23 Feldmeier, A., Kudritzki, R.-P., Palsa, R., Pauldrach, A. W. A., & Puls, J. 1997, A&A, 320, 899 Forman, W., Jones, C., Tananbaum, H., Gursky, H., Kellogg, E., & Giacconi, R. 1973, ApJ, 182, L103 Hatchett, S., & McCray, R. 1977, ApJ, 211, 2 House, L. L. 1969, ApJS, 18, 21 Kaastra, J. S., & Mewe, R. 1993, A&AS, 97, 443 Kahn, S. M., Leutenegger, M., Cottam, J., Rauw, G., Vreux, J. M., den Boggende, T., Mewe, R., & Guedel, M. 2001, A&A, 36, L312 Kallman, T. R., Mukai, K., Schlegel, E. M., & Paerels, F. B. 1996, ApJ, 466, 973 Liedahl, D. A., & Paerels, F. 1996, ApJ, 468, L33 Liedahl, D. A., Wojdowski, P. S., Jimenez-Garate, M. A., & Sako, M. 2001, preprint (astro-ph/010084) Lucy, L. B., & White, R. L. 1980, ApJ, 241, 300 Mewe, R. 1994, in SPEX Handbook of Sample Preparation and Handling, Version 1.0, ed. R. H. Obenauf et al. (4th ed.; Metuchen: SPEX) Nagase, F. 1989, PASJ, 41, 1 Nagase, F., Zylstra, G., Sonobe, T., Kotani, T., Inoue, H., & Woo, J. 1994, ApJ, 436, L1 Owocki, S. P., Castor, J. I., & Rybicki, G. B. 1988, ApJ, 33, 914 Porquet, D., & Dubau, J. 2000, A&AS, 143, 49 Röser, S., Bastian, U., & Wiese, K. 1991, A&AS, 88, 277 Ross, R. R., Fabian, A. C., & Brandt, W. N. 1996, MNRAS, 278, 1082 Sadakane, K., Hirata, R., Jugaku, J., Kondo, Y., Matsuoka, M., Tanaka, Y., & Hammerschlag-Hensberge, G. 198, ApJ, 288, 284

5 No. 1, 2002 SCHULZ ET AL. L2 Sako, M., Liedahl, D. A., Kahn, S. M., & Paerels, F. 1999, ApJ, 2, 921 Sato, N., et al. 1986, PASJ, 38, 731 Schulz, N. S., Canizares, C. R., Huenemoerder, D., & Lee, J. C. 2000, ApJ, 4, L13 Smith, R. K., Brickhouse, N. S., Liedahl, D. A., & Raymond, J. C. 2001, ApJ, 6, L91 Stevens, I. R., & Kallman, T. R. 1990, ApJ, 36, 321 Verner, D. A., Ferland, G. J., Korista, K. T., & Yakovlev, D. G. 1996, ApJ, 46, 487 Waldron, W. L., & Cassinelli, J. P. 2001, ApJ, 48, L4 White, N. E., Swank, J. H., & Holt, S. S. 1983, ApJ, 270, 711 Wojdowski, P. S., Liedahl, D. A., Mauche, C. W., Sako, M., Paerels, F., & Kahn, S. M. 2001, in Two Years of Science with Chandra, ed. M. Garcia (Washington, DC: SAO), 196